Recent research from UC Berkeley reveals that while individual AI models may be deemed safe, their combined use can lead to significant security threats. This study emphasizes that adversaries can exploit the combination of different AI systems using a strategy called task decomposition. This technique involves breaking down a malicious activity into smaller, manageable tasks and assigning them to various AI models based on their capabilities and safety measures. The research demonstrated that such combinations have a significantly higher success rate in producing harmful outputs than individual models alone. For instance, models like Llama 2 70B and Claude 3 Opus, when combined, had a success rate of 43% in generating malicious code, compared to a maximum of 3% when used independently. This finding underscores the escalating risk as AI models improve, highlighting the need for continuous vigilance and red-teaming to mitigate potential misuse throughout the AI lifecycle. The study concludes with a call for persistent scrutiny and experimentation with AI model configurations to identify and address emerging threats.

Source.

TOP STORIES

Nvidia's Leadership Shakeup - What to Expect from Interim CEO Colette Kress
Kress faces tough questions as Nvidia navigates a leadership crisis and scrutiny over its operations …
OpenAI Faces Legal Challenge Over Security Risks and Public Safety
OpenAI is being sued by the California Attorney General over security risks linked to its AI training practices …
Trump Defends Huang Amid Spy Allegations and Nvidia Controversy
Trump believes national security is intact despite allegations against Huang …
AI Tools Empowering Cybercriminals - FBI Issues Urgent Warning
Cybercriminals are using AI tools like Artex to launch swift and devastating cyberattacks …
OpenAI Partners with Safety Evaluators to Enhance AI Security
OpenAI will embed safety evaluators from Apollo Research, Goodfire, METR, and Transluce to enhance AI safety measures …
New AI Tools Launched by OpenAI to Fight Child Exploitation Online
OpenAI introduces new tools to combat child sexual abuse material online …

latest stories